115 research outputs found
Separering - rensing av organisk avfall fra slakterier (system: Reime Econ). Del lI
No full textReime Econs mekanisk -biologiske renseanlegg ble undersøkt for behandling av organisk avfall ved Vest-Ro slakteriet i Sandeid. Anlegget består av et mekanisk trinn som separerer avfallet i en "tørr'' og en "våt" fraksjon, etterfulgt av en kombinert sedimentering og biologisk nedbryting av den flytende fraksjonen. Den "tørre" fraksjonen behandles videre gjennom kompostering, mens renset "våtfraksjon" ledes til kommunalt avløpsnett. I denne prosjektfasen (Fase li) ble anleggets trinnvise og totale renseeffekt målt vha måling av massetransport gjennom anlegget. Videre ble anleggets forbruk av elektrisitet og luft kvantifisert. Gjennom hele anlegget ble avfallets tørrstoffmengde redusert med 91 %, mens hhv 85 %, 67 % og 63 % av organisk stoff (KOF), nitrogen og fosfor i den flytende fraksjonen ble renset fra etter separasjon.
Generelt økt renseeffekten for de enkelte trinn mht tørrstoff og organisk stoff med økende belastning. Derfor var ikke anleggets hydrauliske og stoffmessige kapasitet fullt utnyttet gjennom testperioden. De totale kapital-og driftskostnader for renseanlegget ble beregnet til ca. 184 400 NOK pr. år. I Vest-Ros anlegget tilsvarer kostnadene ca. 37 NOK pr. m3 behandlet avfall (ca. 1500 m3 fast avfall + 3500 m3 vaskevann).Reime Econ a
Zweihundert Hyperbeln auf Herrn Wahl's ungeheure Nase
No full textIn erbauliche hochdeutsche Reime gebracht von Friedrich Hophthalmos [d.i. Johann Christoph Friedrich Haug], der sieben freien Künste Magister. Mit 5 Stahlstichen, gezeichnet und radirt von SonderlandRümann, Arthur: Die illustrierten deutschen Bücher des 19. Jahrhunderts. - 1988Vorlageform des Erscheinungsvermerks: Druck und Verlag von Scheitlin und Zollikofer. - Gedruckt bei J. B. ScherrerStahlsticheIn FrakturRümann 244
Towards Accessible Computational Pathology for Improved Breast Cancer Diagnostics - Exploration, Development, and Integration
No full textBreast cancer (BC) incidence has increased in recent decades. Molecular pathology advances have contributed to the complexity of BC diagnostics, amplifying the workload for each biopsy. Simultaneously, a shortage of pathologists persists. Consequently, there is a need for diagnostic tools that can assist pathologists in their diagnostic works. Digital pathology facilitates analysis of tissue sections on computer screens, enabling remote collaboration, digital annotations, enhanced teaching capabilities, and harnessing the potential of artificial intelligence (AI)-based methods.
Developing AI solutions for BC is challenging. The challenges arise from the large image size of whole slide images (WSIs), the heterogeneity of BC morphology, the absence of large, open, annotated datasets, and the lack of user-friendly standardized software for developing and deploying deep learning models. Training AI models typically requires annotated data, which can be both challenging and time-consuming to generate. Consequently, exploring techniques that mitigate the annotation workload is of interest.
The present thesis consists of three papers. In the first paper, we introduce FastPathology, an open-source and user-friendly software designed for real-time processing and visualization of WSIs on standard computers. FastPathology also generates results that can be easily imported into alternative software for further analysis. Additionally, the software provides the flexibility to incorporate custom deep learning solutions.
In the second paper, a deep learning model for BC segmentation in WSIs was developed. The proposed model, H2G-Net, leverages a cascaded convolutional neural network that refines heatmaps generated by a patch-wise classification model. H2G-Net outperformed single-resolution approaches and enabled rapid inference of WSIs in ~1 second using only the CPU. Furthermore, we introduce a novel hierarchical-balanced sampling scheme, complemented by a custom loss function facilitating end-to-end tissue type balancing. As part of this study, a novel dataset comprising 624 annotated WSIs was curated.
In the third paper, a deep learning model was developed to segment benign and invasive epithelial cells and in situ lesions in WSIs of hematoxylin & eosinstained sections. An immunohistochemistry restaining approach, complemented by pathologists’ annotations, were used to obtain ground truth. Qualitative assessments by pathologists indicated that the best results were achieved for all epithelial cells combined, and for invasive epithelial cells specifically. While the restaining approach reduced the annotation workload, further improvements in the image-toimage registration methods are necessary to minimize downstream task demands.
Throughout this work, we have implemented open, user-friendly solutions allowing clinical users to adopt and test deep learning solutions on their own data. This work also introduces several high performance tools for training, developing, and deploying deep learning solutions for histopathological image analysis
Contributions and challenges of cross-national comparative research in migration, ethnicity and health: insights from a preliminary study of maternal health in Germany, Canada and the UK
No full textBackground: Public health researchers are increasingly encouraged to establish international collaborations and to undertake cross-national comparative studies. To-date relatively few such studies have addressed migration, ethnicity and health, but their number is growing. While it is clear that divergent approaches to such comparative research are emerging, public health researchers have not so far given considered attention to the opportunities and challenges presented by such work. This paper contributes to this debate by drawing on the experience of a recent study focused on maternal health in Canada, Germany and the UK. Discussion: The paper highlights various ways in which cross-national comparative research can potentially enhance the rigour and utility of research into migration, ethnicity and health, including by: forcing researchers to engage in both ideological and methodological critical reflexivity; raising awareness of the socially and historically embedded nature of concepts, methods and generated 'knowledge'; increasing appreciation of the need to situate analyses of health within the wider socio-political setting; helping researchers (and research users) to see familiar issues from new perspectives and find innovative solutions; encouraging researchers to move beyond fixed 'groups' and 'categories' to look at processes of identification, inclusion and exclusion; promoting a multi-level analysis of local, national and global influences on migrant/minority health; and enabling conceptual and methodological development through the exchange of ideas and experience between diverse research teams. At the same time, the paper alerts researchers to potential downsides, including: significant challenges to developing conceptual frameworks that are meaningful across contexts; a tendency to reify concepts and essentialise migrant/minority 'groups' in an effort to harmonize across countries; a danger that analyses are superficial, being restricted to independent country descriptions rather than generating integrated insights; difficulties of balancing the need for meaningful findings at country level and more holistic products; and increased logistical complexity and costs. Summary: In view of these pros and cons, the paper encourages researchers to reflect more on the rationale for, feasibility and likely contribution of proposed cross-national comparative research that engages with migration, ethnicity and health and suggests some principles that could support such reflection
Qualitative evaluation of titanium implant integration into bone by diffraction enhanced imaging
No full textDiffraction enhanced imaging (DEI) uses refraction of x-rays at edges, which allows pronounced visualization of material borders and rejects scattering which often obscures edges and blurs images. Here, the first evidence is presented that, using DEI, a destruction-free evaluation of the quality of integration of metal implants into bone is possible. Experiments were performed in rabbits and sheep with model implants to investigate the option for DEI as a tool in implant research. DE images allow the identification of the quality of ingrowth of bone into the hydroxyapatite layer of the implant. Incomplete integration of the implant with a remaining gap of less than 0.3 mm caused the presence of a highly refractive edge at the implant/bone border. In contrast, implants with bone fully grown onto the surface did not display a refractive signal. Therefore, the refractive signal could be utilized to diagnose implant healing and/or loosening
Treatment of organic waste from animal slaughteries
No full textReime Econ's physical -biological plant for treatment of organic waste was examined at VestRo's animal slaughtery in Sandeid, SW Norway. The system comprises 1) an initial separation of the waste into a solid and a liquid phase, and 2) a combined solid settling and aerobic biological treatment of the liquid phase. The solid phase is further processed by composting, while the treated fluid phase is led to the domestic sewer system. The gradual and overall treatment efficiency of the plant was estimated by monitoring of the mass flux through the system. An overall reduction of 91 % through the plant was found for Dry matter. Wilh regard to Chemical oxygen demand, Nitrogen and Phosphorus of the liquid phase, the removal efficiency was 85 %, 67 % and 63 %, respectively.
The overall treatment efficiency of Dry matter increased with increasing waste loading up to 1,300 kg DM/day. The Bioreactor of the plant was overloaded in the test. However, the capacity can been adjusted in order to achieve an improved treatment efficiency of organic mater (COD). The total fixed and variable costs of the plant was estimated to NOK 185,000 annually (US 4/cu. m).Reime Econ a
- …
